Re: Gcc mainline failed to bootstrap on Linux/ia64

["H. J. Lu" <hjl at lucon dot org>]

On Tue, Sep 09, 2003 at 02:05:21AM +0200, Jan Hubicka wrote:
> > Jan Hubicka <hubicka@ucw.cz> writes:
> > 
> > >> Gcc mainline on Sat Sep  6 07:08:21 PDT 2003 failed to bootstrap on
> > >> Linux/ia64. I got
> > >> 
> > >> In file included from /net/gnu/export/gnu/src/gcc/gcc/libjava/include/jvm.h:23,
> > >>                  from /net/gnu/export/gnu/src/gcc/gcc/libjava/exception.cc:22:
> > >> ./include/java-threads.h:228: internal compiler error: Segmentation fault
> > >> Please submit a full bug report,
> > >> with preprocessed source if appropriate.
> > >> See <URL:http://gcc.gnu.org/bugs.html> for instructions.
> > >
> > > This appears to be related to change of mine to notice_global_symbol.
> > > I just commited fix to different problem that may fix it.  Can you
> > > please ensure that the following patch is in your tree?
> > >        * cgraph.c (cgraph_mark_reachable_node): Only enqueue finalized functions.
> > >        (cgraph_varpool_finalize_decl): Notice global symbol when needed.
> > 
> > This is still not fixed.
> > 
> 
> The problem obviously is that we see global variable placed in register.
> I am quite surprised to see this in java, but the attached patch should
> help.  Can you, please try it out?
> Sorry for the delay - bit too many problems to deal with at once :(
> 
> Tue Sep  9 02:03:42 CEST 2003  Jan Hubicka  <jh@suse.cz>
> 	* varasm.c (notice_global_symbol):  Fix dealing with registers.
> Index: varasm.c
> ===================================================================
> RCS file: /cvs/gcc/gcc/gcc/varasm.c,v
> retrieving revision 1.381
> diff -c -3 -p -r1.381 varasm.c
> *** varasm.c	5 Sep 2003 04:24:26 -0000	1.381
> --- varasm.c	9 Sep 2003 00:03:31 -0000
> *************** notice_global_symbol (tree decl)
> *** 1058,1063 ****
> --- 1058,1066 ----
>         char *name;
>         rtx decl_rtl = DECL_RTL (decl);
>   
> +       if (GET_CODE (decl_rtl) == REG)
> + 	return;
> + 
>         p = (* targetm.strip_name_encoding) (XSTR (XEXP (decl_rtl, 0), 0));
>         name = xstrdup (p);
>   

I can bootstrap gcc mainline with this patch on Linux/ia64 now. But I
got new failures:

+FAIL: gcc.c-torture/compile/20000120-2.c (test for excess errors)
+FAIL: gcc.c-torture/compile/20011119-1.c (test for excess errors)
+FAIL: gcc.c-torture/compile/20011119-1.c (test for excess errors)
+FAIL: gcc.c-torture/compile/20011119-1.c (test for excess errors)
+FAIL: gcc.c-torture/compile/20011119-1.c (test for excess errors)
+FAIL: gcc.c-torture/compile/20011119-2.c (test for excess errors)
+FAIL: gcc.c-torture/compile/20011119-2.c (test for excess errors)
+FAIL: gcc.c-torture/compile/20011119-2.c (test for excess errors)
+FAIL: gcc.c-torture/compile/20011119-2.c (test for excess errors)
+FAIL: gcc.dg/debug/20010207-1.c (test for excess errors)
+FAIL: gcc.dg/debug/20010207-1.c (test for excess errors)
+FAIL: gcc.dg/debug/20010207-1.c (test for excess errors)
+FAIL: g++.dg/init/array11.C (test for excess errors)
+FAIL: libffi.call/closure_fn0.c output pattern test, is 1 2 3 4 127 429 7 8 0
10 11 97 13 19 21 1 3: 756
+FAIL: libffi.call/closure_fn1.c output pattern test, is 1 2 3 4 127 5 6 8 0 10
11 0 13 19 21 1 3: 234
+FAIL: libffi.call/closure_fn2.c output pattern test, is 1 2 3 4 127 5 6 8 0 10
11 0 13 0 21 1 3: 215
+FAIL: libffi.call/closure_fn3.c output pattern test, is 1 2 3 4 5 6 7 8 9 10 0
0 13 0 0 1 3: 72
+FAIL: libffi.call/cls_1_1byte.c execution test
+FAIL: libffi.call/cls_12byte.c execution test
+FAIL: libffi.call/cls_16byte.c execution test
+FAIL: libffi.call/cls_20byte.c execution test
+FAIL: libffi.call/cls_24byte.c execution test
+FAIL: libffi.call/cls_2byte.c execution test
+FAIL: libffi.call/cls_3_1byte.c execution test
+FAIL: libffi.call/cls_3byte1.c execution test
+FAIL: libffi.call/cls_3byte2.c execution test
+FAIL: libffi.call/cls_4_1byte.c execution test
+FAIL: libffi.call/cls_4byte.c execution test
+FAIL: libffi.call/cls_5byte.c execution test
+FAIL: libffi.call/cls_6byte.c execution test
+FAIL: libffi.call/cls_7byte.c execution test
+FAIL: libffi.call/cls_8byte.c execution test
+FAIL: libffi.call/float2.c execution test
+FAIL: libffi.call/float.c execution test
+FAIL: libffi.call/nested_struct1.c execution test
+FAIL: libffi.call/nested_struct.c execution test
+FAIL: libffi.call/problem1.c execution test
+FAIL: libffi.call/pyobjc-tc.c execution test
+FAIL: libffi.call/struct8.c execution test
+FAIL: libffi.special/unwindtest.cc execution test

comparing against 20030903.


H.J.